~ IIIII &II & Ii ' 'Il II I Ill ~ij

PHYSICAL REVIEW B VOLUME 36, NUMBER 4

Phase transitions in quasicrystalline Al6CuMg4

1 AUGUST 1987

M. K. Sanyal and V. C. SahniNuclear Physics Division, Bhabha Atomic Research Centre, Bombay 400085, India

G. K. DeyPhysical Metallurgy Division, Bhabha Atomic Research Centre, Bombay 400085, India

L. VarshneyISOMED Section, Bhabha Atomic Research Centre, Bombay 400085, India

(Received 23 March 1987)

Calorimetric and electron diffraction studies reveal that the quasicrystalline-to-crystalline phasetransition in A16CuMg4 is endothermic, in contrast to the exothermic crystallization in quasicrys-talline Al-Mn reported earlier. This finding indicates that the stability of certain quasicrystallinephases may be much larger than that of the Al-Mn system. This finding is in agreement with theprediction based on quantum structural diagrams. From a kinetic study the activation energy ofthis phase transition was found to be 260 kJ/mole.

Since the discovery of a quasicrystalline phase with afivefold symmetry (point-group symmetry m 35 ) in Al-Mnby Shechtman, Bloch, Gratias, and Cahn' great interesthas been aroused in this class of materials. The quasicrys-talline phase is regarded by some as a new "metastable"phase of solid matter intermediate between crystals andglasses and eN'orts to study properties of quasicrystalshas picked up momentum. In a study of the thermal prop-erties it was observed that transition of the quasicrystal-line Al-Mn system to its orthorhombic crystalline state issimilar to crystallization of metallic glasses. In particular,differential-scanning-calorimetric runs for this transitionrevealed a broad exothermic structure (below 680 K) fol-lowed by a sharp exothermic peak at 780 K.

Recently ' a new class of quasicrystalline alloys hasbeen predicted and produced which has opened up thepossibility of producing large-sized quasicrystals (dimen-sion of a few millimeters) for experimental investigations.The crystalline state of these alloys, which is isostructuralwith the compound Mg32(A1, Zn)49, contains a large num-ber of atoms grouped in clusters associated with each lat-tice point of the bcc unit cell. These clusters which exhibitthe icosahedral symmetry over several states of their for-mation are postulated to provide the "embroyos" for theformation of quasicrystalline phases upon rapidsolidification. A16CuLi3 and A16CuMg4 alloys are two im-portant members in this class of quasicrystals. It hasbeen observed that for A16CuLi3 one can grow largersized single quasicrystals. In this Rapid Communicationwe focus our attention on the other alloy, viz. A16CuMg4and report an endothermic quasicrystalline-to-crystallinetransition in this system. First we will discuss thedifferential-scanning-calorimetric (DSC) and transmis-sion-electron-microscopy (TEM) observations and thenwe will present our finding relating to the kinetics of thistransformation.

The stoichiometric compound (A16CuMg4) wasprepared by melting constituent elements in pure form(99.99%) under argon atmosphere in a graphite crucible.

Quasicrystalline ribbon samples, of about 5-mm widthand 30-pm thickness, for the present study, were obtainedby rapid solidification using the usual melt spinningmethod. The quasicrystalline character of these sampleswas confirmed by TEM observations. Bright field electronmicrograph showing the morphology of the icosahedraldomains and selected area diAraction pattern showingfivefold symmetry are presented in Figs. 1(a) and 1(b),respectively. All other difI'raction patterns were obtainedfrom the same nodule at the expected angles characteris-tic of the quasicrystalline phase. X-ray diAraction pattern(which will be communicated separately) does not showpresence of aluminum peaks. This is in contrast to thecase of rapidly solidified Al-Mn alloys in which the pre-cipitates of quasicrystalline nodules are embedded in analuminum matrix. The ribbons are entirely composed ofgrains of quasicrystalline A16CuMg4, and probably as aconsequence our ribbons are extremely brittle.

Differential scanning calorimetric (DSC) measurementof icosahedral A16CuMg4 samples was performed on aMettler TA3000 thermal analysis system. The systemwas calibrated with pure indium. We conducted the DSC

(b)

FIG. l. (a) The bright-field electron micrograph is displayedshowing the grain morphology and (b) electron diA'raction pat-tern exhibiting fivefold symmetry.

2443 1987 The American Physical Society

2444 M. K. SANYAL, V. C. SAHNI, 6. K. DEY, AND L. VARSHNEY

runs with diferent heating rates. Data collected in thissystem were directly transferred to a modular instrumentscomputer for further processing. The kinetics of transfor-mation was found from these data. Details of all these ex-periments are discussed below.

Two DSC measurements were performed; one on theas-grown quasicrystalline phase and the second on quasi-crystal samples which were heated to 400 C In situ in theDSC setup then cooled to 30 C, before the DSC scanswere made. Above 360 C and below 300'C, the resultsfor both the samples are similar, and both of them show astrong endothermic peak at 474.9'C. However, between300 and 360 C the spectra are very diAerent. In Fig. 2,we have displayed the results for these two measurements,over the range 300-370'C, taken at a heating rate of 30K/min. The symbols I and II in Fig. 2 are identified in thecaption. While for the as-grown quasicrystalline material(I), the heat out vs temperature shows a broad endother-mic structure around 340 C, followed by two sharp peaksat 351.01 and 356.45'C, respectively. (These are desig-nated as A, B, C in Fig. 2.) For the quasicrystalline sam-ples first heated to 400'C (II) and then cooled to 30'Cbefore recording the DSC run from 30 C upwards, thereis only a sharp endothermic peak at 356.45 C coincidingwith the peak of C of II, and neither the broad structure 2nor the peak B is seen. So we believe that A and B areboth associated with the quasicrystalline-to-crystallinetransition. The peak C is connected with a reversible

7. 00

6.00-

CI- 5.00

(polymorphic) crystalline phase transition. This wasconfirmed through the following experiment. The as-grown quasicrystalline samples were first heated in situ to400'C and then DSC runs were performed at a coolingrate of 30 K/min. It was found that the peak C reappears(as an "exothermic" peak) at the same temperatureconfirming the reversibility of this transition. Indeed, thistransition can be reproduced always by going back andforth across the transition point.

To confirm the quasicrystalline-to-crystalline phasetransition, electron diAraction patterns were recorded.These are displayed in Fig. 3. In Fig. 3(a) the threefoldpattern for the on-grown quasicrystalline phase is shown,and in Fig. 3(b) we display the pattern of the as-grownquasicrystal after heat treatment at 340+ 1 C for 1 h. Inthe latter case the sample was removed from the furnaceand allowed to cool to room temperature before taking theelectron diAraction patterns. As can be seen from a com-parison of Figs. 3(a) and 3(b), the separation of spots in

Fig. 3(a) is related to "golden ratio, " while those in Fig.3(b) exhibit a periodicity. This leads to the conclusionthat once the quasicrystalline phase is annealed for 1 h at340 C it transforms to a crystalline state. In order to fur-ther elucidate the nature of peaks 2 and B the as-grownquasicrystalline samples were isothermally kept in DSC atpeak A (340'C) for 1 h. Subsequent DSC scan throughhigher temperature (above 340'C) did not show peak B.In view of this finding peak A was used for studying thekinetics with nonisothermal method, which is discussedbelow.

Study of kinetics of solid-state transformation can becarried out using either isothermal or nonisothermalmethods. In the study of the kinetics of crystallization ofamorphous material using nonisothermal methods, oneseeks a linear relation s between ln(g/Tz) and I/Tz,where T~ is the temperature of the DSC peak correspond-ing to the heating rate Q. One can determine the parame-ters characterizing the devitrification process, namely ac-tivation energy and Arrhenius frequency, from the plot ofln(Q/T~) vs I/T~ (known as Kissinger's plot). We haveapplied this approach to our case of quasicrystalline-to-crystalline phase transition through a study of the shift inposition of the peak 4 of Fig. 2 for five diAerent heatingrates. The experiments were repeated to check the con-sistency of the result; 5 to 7 mg of sample was used ineach run. It was observed that for 10, 15, 30, 45, and 60

U~ 3.00

300.01

1

330.0 360.0

TEMPERATURE( C )

1re

.v

FIG. 2. As recorded DSC data (I) for as-grown quasicrystal-line phase (5 mg of sample) and (II) for in situ annealed quasi-crystalline sample.

(a) (b)

FIG. 3. Electron diff'raction pattern (a) for quasicrystallinephase and (b) for annealed sample as detailed in the text.

PHASE TRANSITIONS IN QUASICRYSTALLINE A16CuMg4 2445

8.75 s

9.25

CV CL 9.75

10.25-

C:

I 10.7 5-

1.51

1.61

17

FIG. 4. Kissinger's plot of quasicrystalline-to-crystallinephase transition (correlation coefficient =0.985).

K/min the positions of peak A were 326.81, 329.78,340.06, 343.26, and 345.83 'C, respectively. Kissinger'splot obtained with this data is shown in Fig. 4. The ac-tivation energy for crystallization and Arrhenius frequen-cy factor were found to be 260.03 kJ/mole and 3.51 && 1022

min, respectively. The value of activation energy ob-

tained here is similar to that observed in the crystalliza-tion of conventional glass (209.7 kJ/mole) (Ref. 8) and inquasicrystalline-to-crystalline transition in Al-Mn(221.76 kJ/mole). The value of Arrhenius frequencyfactor is exceedingly higher than that for conventionalglass (3.41x10 min '), but surprisingly it is close tothat obtained in the crystallization in a gel system (10min ')

We have found that quasicrystalline-to-crystallinephase transition in A16CuMg4 system is an endothermicprocess as compared to exothermic crystallization in Al-Mn. This indicates that A16CuMg4 forms a stabler quasi-crystal than Al-Mn. It is possible that other alloy systemswhich are electronically and compositionally similarwill also show similar behavior. Study is under way toconfirm this prediction. Better stability of these alloy sys-tems was predicted on the basis of quantum structural dia-grams and the result of this Rapid Communicationconfirms that prediction. Endothermic crystallization im-plies that at some temperature below the range of thepresent experiments, the quasicrystalline phase of this ter-nary alloy is more stable than the crystalline phase [iso-structural to (Al, Cu)49Mg32]. This suggests that, by judi-cious choice of cooling rate, it should be possible to growlarge sized quasicrystals of A16CuMg4.

We thank Dr. N. G.S. Gopal for valuable advice onDSC measurements. It is a pleasure to acknowledge valu-able conversations with Dr. S. Banerjee.

'D. Shechtman, I. Bloch, D. Gratias, and J. W. Cahn, Phys.Rev. Lett. 53, 1951 (1984).

~D. Levine and P. J. Steinhardt, Phys. Rev. Lett. 53, 2477(1984).

H. S. Chen, C. H. Chen, A. Inoue, and J. T. Krause, Phys. Rev.B 32, 1940 (1985).

4P. Villars, J. C. Phillips, and H. S. Chen, Phys. Rev. Lett. 57,3085 (1986).

5P. Ramachandrarao and G.V.S. Sastry, Pramana 25, L225(1985).

B. Dobost, J. M. Lang, M. Tanaka, P. Sainfort, and M. Audier,Nature (London) 324, 48 (1986).

H. Yinnon and D. R. Uhlmann, J. Non-Cryst. Solids 54, 253(1983).

SS. P. Mukerjee, J. Non-Cryst. Solids 82, 293 (1986).